Well actually it only goes to 27 kV without the fluorocarbon 114 dielectric.

It’s huge! Shown next to a normal probe:

Just in time for the next run.

This is the same probe used in the Sydney experiment.

Did you go to great lengths to reduce the resistance of the coil path?

He responded:

Yes we try and use a thick enough wire so that the resistance is small but it cannot be too thick in order to be able to have many turns. More importantly is the contact resistance when connecting to the wires. We try to keep this low. From this perspective, a milliOhm meter is helpful.

Milliohm meters start around $200 on ebay. Too much.

Then I found this excellent tutorial on Kelvin (4-wire) resistance measurement.

Luckily I had a current sensing resistor on hand for the current reading:

I used the resistor on the right.

So current sense resistor plus two voltmeters and BINGO: milliohm meter. Results:

The coil path is 0.35Ω.

The dummy coil is 0.185Ω

As a calibration I tested the 0.001Ω resistor. Result: 0.00088888

Not bad.

Now I have excellent tool to troubleshoot resistance in the coil path. AWESOME.

I’ve been running shots on the Polywell yesterday and today.

Just got my deepest potential well yet: 43 Volts.

10KV, 10mA on electron gun. 420V through coils. 8.5 millitorr air:

Be sure to check out the conditions I ran yesterday and today. Each shot has an oscilloscope photo with experimental parameters in caption.

Friday’s photo stream.

Saturday’s photo stream.

Received a bunch of parts from mouser. Including the protection diodes.

Prototyping board for NI USB 6008. I should have ordered two!

Did a tour of my shop neighbor’s guitar workshop:

Made progress on the coil power supply chassis:

And played around with ferrofluid:

• graph the chamber pressure in realtime
• enable/disable fuel
• set the fuel flow in SCCM
• see actual fuel flow rate from mass flow controller
• enable/ disable high voltage
• set frequency and duty cycle for high voltage

The HV duty cycle lets us do fusion trials without melting the grid.

Something I’ve noticed: when the plasma becomes unstable and sparks… it crashes the data acquisition card. I’m guessing the sparks are creating powerful EMFs. We may need shielding.

The replacement PCB is a newer iteration. It bleeds the high voltage automatically. Nice upgrade.

You can read the data from the virtual com port with this command:

screen /dev/tty.usbserial-A800czX2

Each geiger count produces one random bit:

I used google charts to produce these graphs. This code produces the chart from data in mysql. This code transforms the voltage reading to its corresponding instrument value.

I did a quick and dirty downsampling like this:

resamples = samples.in_groups_of(samples.size/number_of_data_points).map{|slice| slice.average rescue 0}

I’m not certain this is all working correctly, but it seems to add up. One oddity: the voltage drops into the negative; perhaps from disabling the high voltage, I’ll have to confirm that.

This is primarily an end to end test of the data acquisition system.

I made headway with the command line program to control the fusor and record data in mysql via ruby. It’s currently setup to record vacuum chamber pressure, voltage and current according to the Glassman, and the effective flow rate of the mass flow controller. Currently I can enter commands to turn the high voltage on/off and set the flowrate of the mass flow controller. Next I want to control the voltage and current on the Glassman.

I also got a geiger counter:

It takes two D batteries. There is a BNC connector for headphones. I have a BNC connector on order with mcmaster. In the meantime I improvised a connection to some computer speakers to test it out. Seems to be working. It picks up the expected background radiation producing that erie clicking sound. You can definitely hear an uptick in the clicks when I run the Fusor.

I’m working towards producing a comprehensive mapping of this device’s performance envelope using computer control to search the parameter space and record the results.

This is all so fun and exciting.

Here is a video:

Details after the jump.

The Sorensen SRL 10-100 allows for remote control of voltage and current. For current control you simply remove the jumper between terminals 7 and 8 and install a resistor to set a current. This model uses 10 ohms per amp – so a 10 ohm resister would set the current to 1 amp. From the manual:

To adapt unit for current mode, resistance programming operation, proceed as follows:

a. Close POWER switch, adjust output voltage to desired compliance setting.

b. Open POWER switch and rotate CURRENT controls fully counter-clockwise.

c. Remove link between terminals TB2—7 and TB2·8, and replace with programming resistor.

CAUTION

Operating unit with open programming leads or links will result in high output current with possible crossover to voltage mode. If ·step changes in output current are to be made by abrupt resistance alterations, use make·before-break switching.

To adapt unit for current mode, resistance programming operation, proceed as follows:

a. Close POWER switch, adjust output voltage to desired compliance setting.

b. Open POWER switch and rotate CURRENT controls fully counter-clockwise.

c. Remove link between terminals TB2—7 and TB2·8, and replace with programming resistor.

CAUTION

Operating unit with open programming leads or links will result in high output current with possible crossover to voltage mode. If ·step changes in output current are to be made by abrupt resistance alterations, use make·before-break switching.

d. Run load to terminals TB3·1 and ‘I‘B3·3, observlng polarity.

e. Set POWER switch to ON. Unit supplies regulated programmed current to the load.

NOTE Any rotatlon of the CURRENT control alters the output current. Dangerous output currents may result. The panel current knobs may be removed lf desired.

So easy enough. But I still need some way to change the resistance from the computer. Enter the digital potentiometer.

This is the specific digital potentiometer chip (MCP4021 – 202E/SN) I used:

Pretty tiny. I had to switch my soldering tip:

Here it is with tiny leads soldered to it:

Very fragile, so I expoxied it to some prototype PCB:

Schematic:

Wired up to the DAQ, and the fluke multimeter set to measure ohms:

Here is an overview of the protocol for wiping the potentiometer:

A 2-wire synchronous serial protocol is used to increment or decrement the digital potentiometer’s wiper terminal. The Increment/Decrement (U/D) protocol utilizes the CS and U/D input pins. Both inputs are tolerant of signals up to 12.5V without damaging the device. The CS pin can differenciate between two high-voltage levels, VIH and VIHH. This enables additional commands without requiring additional input pins. The high-voltage commands (VIHH on the CS pin) are similar to the standard commands, except that they control (enable, disable, …) the state of the non-volatile WiperLock technolgy feature. The simple U/D protocol uses the state of the U/D pin at the falling edge of the CS pin to determine if Increment or Decrement mode is desired. Subsequent rising edges of the U/D pin move the wiper. The wiper value will not underflow or overflow. The new wiper setting can be saved to EEPROM, if desired, by selecting the state of the U/D pin during the rising edge of the CS pin.

So I wrote a small ruby program using the DAQmx library to slowly ramp up and down.

Short circuit the power supply for current mode operation:

Finally hook up the assembly to the power supply:

SUCCESS as you can see from the video at the top of the post!

A new mass flow controller costs roughly $1,100. There are plenty available on ebay starting from $70.

A mass flow controller must be calibrated for a specific gas. Some can be calibrated on demand with a digital interface, and some are hard wired to work with a specific gas. I’m wondering if we can get away with using a hydrogen calibration for our deuterium system?

The next parameter is flow rate. Here is some background on flow rate nomenclature: flow rate nomenclature. Although mass flow controllers meter based on mass flow rate, they confusingly are rated using volumetric flow rate metrics like SLPM (standard liters per minute), SCCM (standard cubic centimeters per minute) or SCFH (standard cubic feet per hour). Conversion between the two must be done at standard temperature and pressure, and must take into consideration the density of the gas: details.

So before I can spec a mass flow controller I need to calculate our midpoint flow rate needs.

Some facts:

The chamber holds ~6,500 cm^3

The turbo pump removes gas at the following rates:

N2 -> 56 liters/sec

He -> 48  liters/sec

H2 -> 36  liters/sec

So the mass flow controller meters how much fuel is flowing into the chamber. We also want to control how much fuel the vacuum pump is pulling out of the chamber. This requires a gate valve or butterfly valve. Many of them include an integrated bellows connection (which is great), and are pneumatically actuated (not so great).

Adding a gate valve will require us to redesign the welded sled; and I bet it won’t be the last design change. To keep it flexible I’m going to take another page from Andrew and switch to 80/20.

Now my naive attempt to calculate flow rate needs:

Without a gate valve the pumps will remove 2160 l/min. Now the typical mass flow controller seems to max out at 10 SLM (l/min)…. so the pump will remove fuel far faster than the mass flow controller can supply it. And this will waste expensive fuel.

I think these calculation work for pressures in the laminar flow region (> 1e^3 torr). We are working with pressures between the laminar flow domain and the free molecular flow domain – I don’t know how the calculations change at this point.

ps. Wolfram Alpha seems to be useful for density calculations.